AI coding agents can accidentally skip important safety checks when they commit code. They might use flags like --no-verify to bypass pre-commit hooks. That means linting, formatting, and security tests are ignored.
This hook stops those bypass flags before the command runs. It checks every bash command for dangerous patterns. If found, it rejects the command and shows a clear message explaining why.
Teams using Claude Code can enforce commit quality gates with this setup. Every commit must go through pre-commit hooks. Code stays clean, tested, and secure.
Global
mkdir -p ~/.claude/skills/block-no-verify-hookProject
mkdir -p .claude/skills/block-no-verify-hookSource Repository
Caveman Commitjuliusbrussee/caveman
Generate ultra-compressed commit messages with Conventional Commits format
Verification Before Completionobra/superpowers
Always run fresh verification before claiming work is complete or passing
Git Guardrails Claude Codemattpocock/skills
Stop dangerous git commands from running automatically with safety hooks
Setup Pre Commitmattpocock/skills
Automate code formatting type checking and tests before every commit
Using Git Worktreesobra/superpowers
Isolate your feature work with git worktrees and keep your main branch safe
Finishing A Development Branchobra/superpowers
Finish your development branch with merge, pull request, or cleanup options
Resolving Merge Conflictsmattpocock/skills
Resolve Git merge conflicts step by step preserving both intents
Git Commitgithub/awesome-copilot
Automatically write standardized git commits by analyzing your code changes